We developed the Tozan platform to improve how companies test and learn, utilizing AI to make experimentation more efficient and valuable. There are important technical improvements over how traditional A|B testing works across industries, but this article discusses an important conceptual, not technical difference.
As mentioned in prior posts, Tozan solves three core challenges in traditional A|B testing — 1) measurable waste 2) arbitrary versions and 3) moving target.
But there is a single, common thread across many A|B tests that serve as an umbrella to inefficiencies following therefrom: curiosity. Many A|B tests start and end with curiosity — how will version 1 perform relative to version 2, why did user segment ABC respond more positively to version 3, what caused version 4 to outperform version 3 among user segment XYZ? Deliberations like these arise naturally and in some respect service important thinking around business and product strategies. However, within the context of experimentation, being ruthlessly goal and action oriented pays dividends. What is the goal? The goal of any organization is to optimize its Key Performance Indicator(s). The KPI is the reductionist approach to strategy and alignment. Many experiments, riddled with curiosity, descend into labyrinths of data pivoting, confusion, and endless chains of successive tests with no benefit to the business. These experiments have the common trait of following interesting questions and quickly lose the connection to KPI optimization. It is easy to become intoxicated with the volumes of user generated data, but it’s important for experimentation, and more broadly Big Data, to be used as a tool, not as vice.
Tozan does not aim to answer the many questions that arise from Big Data and experimentation. The purpose of the Tozan platform is to enable fast and efficient KPI optimization within an experiment. What matters most to Tozan is the growth of the organizational KPI(s), not why they grew, or what variable interactions caused a given segment of users to take a specific action. If a variant or set of variants works to maximize the KPI, Tozan promotes them to the surface without expounding upon the reasoning and causality. The most essential ingredient that a company needs to have to work with Tozan is an agreed upon KPI for the experiment and, ideally, a clear map of how that KPI relates to the organizational KPI (if they are not one and the same). If your company does not have that level of detail in its KPIs, then Tozan can help get you started there too.
